JP2002523964A - Processing for simultaneously measuring propagation characteristics of multiple radio frequency channels - Google Patents

Abstract

A process for simultaneously measuring the propagation characteristics of a plurality of radio-frequency channels.According to the invention, from several base stations are transmitted signals whose spectra are interleaved. At the receive end, each of these spectra may be reproduced and the impulse response of each channel may be calculated.Application to radio-communications with mobiles.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

TECHNICAL FIELD OF THE INVENTION

An object of the present invention is a process for simultaneously measuring the propagation characteristics of a plurality of radio frequency channels. The invention finds application in automotive radio communication. The radio frequency channel whose characteristics are measured is broadband here.

[0002]

2. Description of the Related Art

2. Description of the Related Art In an automobile wireless communication system, for example, a GSM system or IS-9
As in the five system, a microwave signal propagates between a fixed location base station connected to the telephone network and a car user, hereinafter referred to as a car station. Each connection between a base station and an automobile station is made by radio frequency or an automobile radio channel.

[0003] Microwave signals reach mobile stations (in the so-called down direction) and base stations (in the so-called up method) with slightly attenuated and out-of-phase echoes, which are encountered during propagation. This is the result of the obstacle. The signal experiences an overall attenuation that varies with the movement of the mobile. The two movement criteria are distinguishable. High speed fluctuations caused by jamming and low speed fluctuations due to changes in the environment when mobiles are moved long distances and in the separation of mobile station based stations. These attenuations first affect the signal-to-noise ratio. However, multipath also creates intersymbol interference, which is indicative of a faster information rate.

In order to improve the performance of the transmission, it is conceivable to transmit from several different sources (transmitter diversity) or to receive at different locations (receiver diversity). Automatic transmission (or "handover") procedures are used when one of the connections is highly attenuated or heavily subject to inter-symbol interference. Processing algorithms may also combine signals from several channels.

[0005] As high bit rate transmission systems become faster, spatial or polarization diversity is obtained. Spatial microdiversity (using several close transmit and receive antennas) or polarization microdiversity (two transmit and receive polarizations) allows to counter multiple paths and fast fading. Macro diversity (connection between a mobile and some base stations) can also be used to terminate it. Further, the transition between cells can be smoothed against the mask effect.

[0006] It is therefore essential to have a general control over the measurement of the propagation characteristics of the different channels, which can be used in diversity so that new values can be designed and arranged in an optimal way.

The measurement period for macro diversity at the mobile station level or for polarization diversity at the mobile station level does not pose any particular problem. The channel detection signal is transmitted from the base station. The antenna is placed on a mobile with a receiving means,
These antennas are separated by a few centimeters in spatial microdiversity or superimposed in polarization diversity. They are connected to the same sensing device, which, for example, alternately reads both signals provided by the antenna. Rotating the wheels of the vehicle causes signal capture. It can be said that the configuration is “main distance”.
A close examination of the results makes it easy to identify the exact location of each measurement point.

[0008] Microdiversity and macrodiversity at the base station level are very difficult to obtain. These difficulties can be emphasized by distinguishing between the "main time" method and the "main distance" method.

A) In spatial microdiversity in a “main time” method base station, some antennas have several c
m. In polarization microdiversity, the antennas are overlaid.

[0010] The channel between the mobile station and the base station is complementary, and the signal can be transmitted very well from the mobile station and received with the diversity of the base station. The receiving end technician must be present at all times to start and stop the measurement. Signal acquisition occurs at a normal rate and they are controlled by the clock at the receiving end. This method is called “main time”.

With this configuration, the vehicle must move at a constant speed so that the exact position of the measurement point can be restored. This is only possible for small sections, and some routes must be excluded for road traffic.

In macro diversity, antennas are located several hundred meters apart.
Channel measurement at the "main time" therefore requires several sensing devices when something happens at each station. Professional technicians must be present at each location. The issues raised regarding microdiversity remain. In addition to these issues,
There is a problem in synchronous measurement, the time to wake up and stop recording must be common for different locations.

B) "Main distance" method Different methods have been proposed for performing measurements with diversity at the base station or macro diversity on two channels and in a "main distance" type configuration.

The first solution, which is the most conceptually straightforward, is to pass the same sensing signal through two connections. A superposition of the impulse responses of the two channels is obtained. If deviated sufficiently in time, these responses are separated. For this, the sequence of the transmitter has to be synchronized. This technology is described in M. G
. KADEL, 1994, Stockholm, Sweden, IEEE, proc.
of VTC, “Measurement of wideband micro- and macro-diversity characterist
ics of the mobil radio channel ", pages 165-169.

[0015] Synchronization of the sequences is possible a priori in microdiversity or polarization diversity. Implementing in macro diversity is complicated because two independent transmitters work together. The sequence can always be shifted arbitrarily by initializing the sequence of any one of the transmitters. In any case, good transmitter synchronization is only confirmed at the receiving end by displaying the main impulse response peaks and paying attention to their effective separation. It is only in this situation that the measurement of the selection is started.

The length of the sequence must be at least equal to twice the delay spread of the measured impulse response. In macro diversity, the peaks of the two impulse responses alternate independently of one another during the movement of the mobile. Therefore, there needs to be enough space to move, especially when re-initialization of the transmitter can arbitrarily achieve synchronization.

The sequence length used in this method is on the order of 100 μs.
This method is difficult to apply to more than two transmitters, since sequence identification and synchronization will prove to be difficult to perform.

Another problem with macro diversity arises from the fact that the oscillators of the two transmitters fluctuate separately. They are not due to the same clock, and the calculated impulse response may be in the same end overlap. The result is
Unfortunately, these fluctuations are possible because they become unusable and difficult to confirm stability during the measurement time if they actually occur.

As a second solution, G. Written by KADEL, Ottawa, Universal Personal Commu
"Simulation of the DECT Sys", a bulletin of the second international conference of nications (ICUPC)
This solution consists in shifting the carrier frequency of the second transmitter to that of the first. In practice, about 20 Hertz is used. The offset Δf results: the received composite signal is demodulated to the frequency of the first transmitter, and after processing, two overlapping impulse responses will be observed. A large frequency shift leads to an artificial Doppler effect, since the measurement remains good, since for most 100 μs, the sequence is very short compared to the 50 ms of the Doppler effect. When the mobile is in sleep mode, the impulse response associated with the second channel fluctuates irregularly in time, whereas the first one does not. Two methods can be considered.

For each measurement point, the moving body stops and starts a series of recording. After processing, the Doppler spectrum of each delay is calculated at the baseband from successive responses. Contributing to this spectrum of channel 1 is, in principle, one row of zero-frequency lines, that of channel 2, a line separated by Δf from the first. The law of sampling is satisfied when acquisition occurs every 25 ms. A low-pass filter in baseband blocks information from any one channel.

When the moving object is moving very slowly, it is considered that the artificial Doppler effect precedes the actual Doppler effect. The capture is then considered to take place on the move and always at the correct time intervals. The contribution of each channel to the Doppler spectrum for a delay is no longer a simple line. However, they continue to concentrate around this line. As a precaution, in practice, the capture should be less than 25 ms. 2
To prevent the two frequencies from overlapping, the speed of the mobile is limited to 1 m / s (4 km / h).

This method also has many disadvantages. First, we do not propose an actual "principal distance" method. In practice, it is forced to stop at each measurement point in the first case, or to proceed very slowly (4 km / h) at a constant speed in the second case. In addition, channel separation slows down the computation. Each impulse response is obtained by filtering the information observed over approximately one hundred consecutive acquisitions at some delay. This process must be repeated for each sample of the received sequence. The processing complexity of the receiving end is increased by at least 10,000 factors that are at least related to diversity-independent measures.

If measurements occur every 25 ms, the time required to capture about 100 sequences is 2 seconds or more. Therefore, the channel is required to be fixed over this time. Such a hypothesis is not well validated when a vehicle moves around a mobile.

The purpose of the present invention is indeed intended to overcome these disadvantages of the prior art. It proposes a "main distance" type process that allows simultaneous analysis of several channels in the same band, while combining ease of execution with signal processing simplicity. The present invention
Performed to observe the impulse response over a time T that is compatible with the environment being sensed.

[0025]

[Means for Solving the Problems]

In the process of the present invention, one sequence of the transmitter does not have to be reinitialized, or the deviation between the sequences does not have to be followed on time. The processing is performed in real time, and by measuring at a certain time, the impulse response at that time can be calculated. The computation time is very economical and its complexity is that of a measurement technique that is not bound by diversity and is multiplied by the number of channels.

The measurements are controlled at a distance and the vehicle is no longer forced to move at a constant speed. Thus, reconfiguring the journey is not a major problem. You no longer need to mobilize technicians to make measurements at each site. Motivational problems arising with macro diversity can also be avoided.

Precisely, an object of the invention is to provide simultaneous p radio frequency channel propagation characteristics (an integer p equal to at least 2) between p base stations and mobile stations over an observation period T. Is a measurement process. The present application provides:-a periodic signal with a period pT is transmitted simultaneously from each base station and said signal transmitted by a base station of rank n (n is between 1 and p), where k is the number of lines (
k / T) + ((n-1) / pT) has a spectrum composed of lines placed at a frequency of -p signals transmitted by p base stations are received simultaneously by the mobile station , Wherein the received signal is processed in a time window of width pT to extract p impulse responses of p radio frequency channels.

To transmit a signal of interest from the nth station, the following procedure is taken. A sequence of elements of period T is produced; this sequence is produced with period T so as to obtain a periodic signal of period T; a carrier having a frequency equal to the center frequency Fc of the frequency band to be analyzed. And the carrier frequency is shifted by an amount of (n-1) / pT, and the resulting signal is transmitted after modulation of the carrier so offset.

Preferably, the band of the transmitted signal is reduced by filtering to limit it to a band of width B, and the spectrum of the transmitted signal is −B / 2 (
Inclusive) to B / 2 (inclusive), so the number k takes on all integer values between -N (inclusive) and N-1 (inclusive), where N is Equal to BT / 2 (
This seems to be an integer).

In order to obtain the signal transmitted by the base station, a memory can also be loaded together with a suitable sample of the signal having the spectrum of interest.

As far as the processing performed in the mobile station is concerned, preferably the received signal is
It is sampled at a sampling frequency at least equal to the width B, which is the band used for transmission.

To obtain the impulse response of each channel, the following procedure is chosen. a) The decision is made from: The amplitude of the line on the frequency of k / T for obtaining the first spectrum; the amplitude on the line of the frequency of k / T + (n-1) pT for obtaining the n-th spectrum; and the k for obtaining the p-th spectrum. B) For each spectrum, the ratio of the amplitude of that line to the amplitude of the corresponding line of the transmitted signal is calculated. C) Obtain for each spectrum the frequency of / T + (p-1) pT. A different ratio inverse Fourier transform is performed.

Another possible procedure is to correlate the received signal with the different signals transmitted by the first to nth over the window of pT, which is the first This is to obtain the nth to nth impulse responses through the pth channel.

The process just defined applies to any number of channels. As a special case, if this number is equal to 2, the process is as follows. The signal transmitted by one of the stations has a line spectrum of frequency k / T. The signal transmitted by the other station has a line spectrum of frequency k / T + (1 / 2T). The signal received by the mobile station is processed in a window of width 2T.

[0035]

BEST MODE FOR CARRYING OUT THE INVENTION

 FIG. 1 shows the SB₁From SB_nThrough the SB_pUp to p base stations up to
1 shows a wireless communication system including a mobile station SM. C₁To C_nThrough C _p Up to p radio frequency channels are thus established between the base station and the mobile station.
Is defined by The present invention measures the impulse response of each of these p channels.
And do so at the same time.

Before discussing this approach, it may be worthwhile to recall the general principles of signal processing in the case of telecommunications.

The radio frequency channel will be considered as a linear filter. It is known that such a filter can be defined as an impulse response, which is the response of the filter to an input signal that would be a Dirac pulse. The concept of an impulse response is particularly useful because the filter response y (t) can be calculated for any input signal x (t). by this,

(Equation 1) Here, h (t) is an impulse response, and (star) indicates a product of convolution.

If X (f), X (f), and H (f) represent Fourier transforms of x (t), y (t), and h (t), respectively, the following equivalent relations are obtained. Is given.

(Equation 2) This can be expressed in frequency when X is further canceled.

(Equation 3)

In the hypothesis that x has a good autocorrelation property, equation (1) is further expressed as follows.

(Equation 4) here,

(Equation 5) Where the minus sign represents a complex junction and the symbol δ represents a Dirac distribution.

Thus, the impulse response h (t) can be obtained from the radio frequency channel in two ways. Knowing the spectra X (f) and Y (f), the ratio of Y (f) / X (f) is calculated, which gives H (f) according to equation (3). H (t) is found by the inverse Fourier transform. -Convolution between transmitted and received signals

(Equation 6) Where h (t) is directly given by equation (4).

The first method is known as the “frequency” method (otherwise the inverse, with the implication of the Fourier transform), and the second method is known as the “time” method.

FIGS. 2A, B, 3 A, B, and 4 show some signals and some spectral velocities in this technique and illustrate these two methods. First, FIG. 2A shows, by way of example, a bit sequence x (t) that takes on values from +1 to -1 over a period of 500 μs. FIG. 2B shows the spectrum X (f) after filtering. It extends above about 30 MHz. If such a signal modulates a carrier (not shown), the signal y (t) shown in FIG. 3A
Is received and modulated. The corresponding spectrum Y (f) is shown in FIG. 3B.

To obtain the impulse response h (t) of the channel that transmitted this signal, it is therefore possible to proceed in the two different ways described above. By the frequency method: the spectrum X (f) of the transmitted signal and the spectrum Y (f) of the received signal are available, the ratio of Y (f) / X (f) is calculated and then the inverse Fourier transform The impulse response is again obtained. By the time method: knowing the received signal y (t) and the transmitted signal x (t),
Product of convolution

(Equation 7) Which directly gives the impulse response for h (t).

With all this in mind, let's return to the invention itself. The special case of two channels is described, but will soon increase to any number of channels.

When operating in a particular mode, a bit sequence for a period T is first created like the sequence shown in the example of FIG. Bit to +1 or -1
However, a sequence of 1 and 0 may be used. The period T is several tens of microseconds, for example, 20 μs.

The sequence of the period T is similarly repeated. Thus, a periodic signal of time T is obtained, the two periods of which are shown in FIG.

Thereafter, a carrier is created, the speed of which is shown in FIG. 6A. Transport OP ₁ has a certain frequency f _0. This value is an example of 2.2 GHz. The time scale of FIG. 6A is therefore not relevant to the time scale of FIG. This carrier is offset in frequency by an amount １ ／ T so as to constitute the second carrier OP ₂ (FIG. 6B). After a time T, this second carrier OP ₂ is in phase opposition to the first carrier OP ₁ , but returns to that phase with a spacing of 2T.

The periodic signal of FIG. 5 is modulated together at the same time as the first transport OP ₁ and the second transport OP ₂ . The obtained signals s ₁ (t) and s ₂ (t) have a period T at the beginning and a period 2T at the second.

The baseband spectrum (ie, around the center frequency f ₀ ) is shown in FIGS. 7A and 7B. Since the signal of FIG. 5 has a period of time T, the spectrum S ₁ (f) corresponding to the carrier OP ₁ consists of a line of frequency k / T, where k is an integer. With respect to the transport OP _2, spectrum S ₂ (f), although there is a frequency offset of 1 / 2T, it is constituted by the still k / T apart lines. The line is therefore located at (k / T) + (1 / 2T).

The spectra S ₁ (f) and S ₂ (f) are restricted to a band of value B, for example, by transmitting end filtering. Thereafter, they are changed from -B / 2 (inclusive) to B / 2 (
Do not include). (In fact, the window of the hamming frequency is given to the received signal). Assuming that N / T = B / 2, the number k takes a total value between -N and N-1 (inclusive).

[0051] The spectra of the two signals transmitted by the two base stations are therefore interleaved.

If one of the periodic signals is multiplied by the other signal and the sum consists of the product obtained over a period of 2T (the least common multiple of that period), the opposite amount is the period T Obtained within two time periods, resulting in a sum of 2T of zero. Thus, the two transmitted signals can be said to be "quadrature."

It will be appreciated that in accordance with the present invention, each signal transmitted by a base station has a characteristic in the spectrum of events that is distinguishable from signals transmitted by other stations. In addition, the transmitted signals are orthogonal to each other, so they do not mix and maintain individual characteristics. Thus, propagation occurs through two channels.

FIGS. 8 and 9 show the configuration and two methods of transmission of signals s ₁ (t) and s ₂ (t) by two base stations. In FIG. 8, first, two combiners S ₁ and S ₂ are connected to the carrier O
Distribute the P ₁ and OP _2, the second combiner is seen that are offset frequency only 1 / 2T from the first combiner. It can also be seen that there is a pseudo-random sequence generator G and two multipliers M ₁ and M ₂ that receive two carriers and a sequence.
These multipliers distribute two signals s ₁ (t) and s ₂ (t). Finally, the two transmitters E ₁ , E ₂ provide the appropriate power and amplify these signals so that the two antennas A ₁ , A ₂ transmit radio frequency waves towards the mobile station.

FIG. 9 shows the signal s₁(T) and s_TwoThe sample of (t) has two memories M₁, M_TwoTo enter the
, They show a variant read at clock speed H. Two antennas A ₁ , A_TwoTransmitters E connected to₁, E_TwoAlso allows radio frequency transmission.

The signals s ₁ (t) and s ₂ (t) so transmitted are received at the mobile station and continue to be processed to regenerate the impulse response of each channel. As already indicated, the received signal is observed within a time window of period 2T. It is known that frequency filtering occurs according to a function of the type sin (πfτ) / πfτ due to time constraints of the signal, where τ = 2T. This function is known as "cardinal sine". Each signal line so limited is developed in the frequency direction as shown in FIG. 10A for the line in the middle of the band, which is the center of the signal s ₁ (t) transmitted by the first station. This is the case for the line. FIG. 10B shows the same event for a second signal line that is placed at １ ／ T. It can be seen that one major protrusion of the cardinal sign coincides with the other zero and vice versa. This interleaving occurs for all lines of the two signals with the result that the spectra of the two signals do not mix at the receiving end.

FIGS. 11A to 11D show some signals included in the frequency processing mode. First, FIG. 11A shows the spectrum at the transmitting end of the first signal, s ₁ (f), which will be received if the transmitter and receiver are connected by an ideal cable. It is the spectrum of the signal that would be. FIG. 11B shows the actually received spectrum R ₁ (f). The function H (f) in FIG. 11C is given by the complex amplitude ratio of the line. An inverse Fourier transform (eg, inverse DFT) can return to the time of obtaining the first channel impulse response h (t), which is shown in FIG. 11D.

Integrally expressed as follows. h ₁ (t) = DFT ⁻¹ [R ₁ (k / T) / S ₁ (k / T)] Similarly, for the second channel, the calculation is as follows. h ₂ (t) = DFT ⁻¹ [R ₂ ((k / T) + (１ ／ T) / S ₂ ((k / T + ／ T)]]

Instead of using the distance Fourier transform DFT, a fast Fourier transform (FFT), which has a fast calculation time, can be used. Implementation is facilitated when the total number N associated with the number p of transmitters without lines is a power of two. By adjusting the parameters B and T, an appropriate integer N can be obtained. If the number p of base stations is not a power of 2, the integer p 'is determined to be a power of 2 which is larger than p and closest to p. It will be appreciated that the method described above is given this new value, which is virtual, with p'-p base stations distributing no signals.

For example, in the case of six base stations, two virtual stations are considered so that the number of stations is eight, and an exact power of 2 and all processing are performed as if eight (ie, 2 ³ ) It must be remembered that the stations are implemented as if they are, and that the real two of these stations are virtual.

The variant just described relates to the type of frequency. Variations in time can also be implemented by providing a correlation between the transmitted signal and the received signal. The received signal passes through a bank of two digital filters adapted to the pattern (of duration 2T) transmitted from each channel. The output of the filter clearly gives two consecutive copies of the transmit channel impulse response (transmit-propagate-receive). The first copy is out of phase with the expected impulse response, and the second copy is out of phase by the amount -π. By omitting, an impulse response that is expected to be sampled at the frequency B during the period T is extracted.

If there are p channels instead of two channels, a more general way of processing channels of rank n (1 <n <p) can be selected from: -Only 2N information lines are left at the frequency (k / T) + (n-1) pT. The corresponding line is centered at frequency (n-1) / pT. A complex amplitude line,
Calculate the ratio of the corresponding signal to the complex amplitude line received when the transmitter and receiver are connected by an ideal cable. The time is traced back by the inverse DFT and then corrected by the quantity e ^{j2 π (n-1) / pT} . The estimate of the impulse response of the nth propagation channel sampled at frequency B during time period T is thus recovered. Keep all 2Np lines and use a value of zero instead of 2N (p-1) lines that are not useful. It is possible to observe p consecutive copies of the nth impulse response per phase shift period of the total pT period.
The first copy is not out of phase with the expected impulse response, the kth copy is out of phase by the amount {−2π (n−1) (k−1)} / p, and finally The p-th one is out of phase by the amount {-2π (n-1) (p-1)} / p. By omitting, an impulse response estimated over the period T and sampled at the frequency B is extracted.

In the time method, the received signal passes through a bank of p digital filters of the pattern (of duration pT) transmitted from each channel. Around one phase shift period, the output of the nth filter obviously gives p consecutive copies of the nth transmit channel impulse response (transmit-propagate-receive). The first copy is one with the expected impulse response out of phase, and the n-th is -2π (n-
1) out of phase by the amount of / pT, and finally the p-th is -2π (p-1
) Is out of phase. By omitting, the impulse response estimated over the period T and sampled at the frequency B is extracted.

For the sake of explanation, an experiment performed with macro diversity by two transmitters will be described.

The first source periodically sends a sequence of 255 bits at a flow rate of 12.5 Mbit / s. This signal modulates a carrier on a frequency of 2.2 GHz, which is then filtered and amplified. The analyzed band is 2.2 GHz
, And exists at the maximum 25 MHz when N = 255. The value of the period T is 2
0.4 μs. Therefore, the value of the frequency offset of channel 2 is ＴT-
24.451 kHz. Incidentally, it will be seen that this offset is very large compared to about 20 Hertz seen in the prior art after introducing the artificial Doppler effect. The frequency offset is calculated by using each frequency f ₀ using a synthesizer.
And f ₀ + ／ T (FIG. 8) during the modulation period. In addition, πt / 2
By out-of-phase by T, a sequence stored in baseband by the second transmitter over a period of 2T can be obtained. This sequence, which is read by modulating the frequency carrier f ₀ having a period of 2T with an amplitude, gives an expected signal. In this configuration, the combiners both deliver the same frequency f ₀ .

In a first example, two radio frequency sources are co-located and they are visible to the mobile station. After processing, two identical (and offset) responses are obtained, confirming the experiment. This is shown in FIG. The transmitters are then placed on the roof and spaced 600 meters apart, which is a typical macro diversity configuration in small urban cells. The effective power transmitted from each source is 40 dBm. The receiver is located on the mobile vehicle. Acquisition is 4
It lasts 0.6 μs and contains 1020 samples. Measurement points are about 2
cm and the path is 60 meters long. One set of examples of the impulse response measured in this process is shown in FIG. This demonstrates the excellent power of the calculated response.

The receiving device comprises an automatic gain control circuit for adapting the power of the received signal to the electronics of the receiver. It is desirable that the signals from the two connections reach the receiver at comparable average power levels to prevent the channel distortion from becoming too large. This is obtained by performing a measurement of the average narrow-band field in advance, and subsequently adjusting the signal strength. This caution is even more effective whatever the "main distance" from the diversity type measurement.

A bias on one carrier frequency of the transmitter with respect to the nominal value or deviation of the flow rate at one transmitting end of the channel can change the performance of the link. These two drawbacks cause a fundamental sine overlap at a frequency of k / 2T, which was investigated by simulation. A carrier frequency error of about 100 Hertz or more, ie, a hundredth order of line separation, and an inaccuracy of the hundredth order, ie, a hundredth order between samplers, over a period T of about 20 nanoseconds or more. The performance degradation for is acceptable. These tolerances are fully considered by the available hardware.

[Brief description of the drawings]

FIG. 1 illustrates a wireless communication system with several base stations and one mobile station.

FIG. 2 shows the corresponding spectrum after bit sequence and filtering.

FIG. 3 shows the reception sequence and the corresponding spectrum after filtering.

FIG. 4 shows the speed of an impulse response.

FIG. 5 shows an example of a bit sequence for creating a periodic signal.

FIG. 6A shows a first carrier, and FIG. 6B shows a second carrier obtained from the first carrier when the frequency offset is ＴT.

FIG. 7A shows the spectrum S ₁ (f) of the _first signal transmitted by the first base station, and FIG. 7B shows the line of the second signal transmitted by the second base station. The spectrum S ₂ (f) is shown.

FIG. 8 shows a first embodiment of a means capable of generating two signals transmitted by a base station.

FIG. 9 shows a second embodiment of the means capable of generating two signals transmitted by the base station.

FIGS. 10A and 10B show the influence of a window having a width of 2T applied to a signal.

FIGS. 11A, 11B, 11C and 11D show different signals mainly appearing in the case of several signal processing.

FIG. 12 shows an example of two impulse responses in a situation where the base station is visible from the mobile station.

FIG. 13 shows examples of two other impulse responses in a situation where the base station is not visible from the mobile station.

[Explanation of symbols]

SB ₁ ,. . SB _n ,. . SB _p ... base station SM ... mobile station E _1, E ₂ ... transmitter S _1, S ₂ ... synthesizer M _1, M ₂ ... memory G ... pseudorandom sequence generator

──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5K004 AA01 BA02 BB04 5K042 AA06 CA02 CA23 DA19 EA15 FA11 5K067 AA33 CC24 EE02 EE10 EE24 LL11

Claims (9)

[Claims] 1. A plurality of p base stations and p mobile stations over an observation period T
Simultaneously measuring the radio frequency channel propagation characteristics (an integer p equal to at least 2) comprising: transmitting a periodic signal of period pT from each base station simultaneously, and of rank n (where n is 1 to p) The signal transmitted by the base station is (where k is the number of lines).
k / T) + ((n-1) / pT) has a spectrum composed of lines placed at a frequency of -p signals transmitted by p base stations are received simultaneously by the mobile station , The received signal is processed within a time window of pT width to extract p impulse responses of p radio frequency channels. 2. To transmit at the level of the nth station, a sequence of elements of period T is produced, this sequence is produced at period T so as to obtain a periodic signal of period T, A carrier is created having a frequency equal to the center frequency F ₀ of the frequency band to be analyzed, the frequency of the carrier is shifted by an amount of (n−1) / pT, and the resulting signal is The process according to claim 1, wherein the transmission is performed after modulation of the offset carrier. 3. The band of the transmitted signal is reduced by filtering to limit it to a band of width B, and the spectrum of the transmitted signal is -B /
2 (inclusive) to B / 2 (inclusive), so the number k is -N (inclusive)
The process according to claim 2, wherein all integer values between N and (not including) are taken, where N is equal to BT / 2. 4. The memory according to claim 1, wherein the memory is read together with samples of the signal having the spectrum of interest in order to obtain the signal transmitted by the base station. Processing. 5. The process according to claim 3, wherein before the p signals received by the mobile station are processed, the received signals previously demodulated to the frequency F ₀ are sampled. . 6. In the mobile station, the processing includes: a) obtaining (p) spectra (R _n (f)) by using a frequency (k / T) + ((n
-1) determining the amplitude of the line on / pT); b) calculating, for each spectrum, the ratio of the amplitude of the line to the amplitude of the corresponding line of the transmitted signal; c. The processing according to claim 5, further comprising :) performing an inverse Fourier transform of another obtained ratio. 7. At the mobile station, the processing consists in correlating the received signal with another signal transmitted by p base stations, which in this way comprises the p channels. The process according to claim 5, wherein each impulse response is obtained. 8. The received, demodulated and sampled signal passes through a bank of p digital filters adapted to the periodic signal of period pT, and at the output of the nth filter the nth P of the impulse response of the channel
8. A process according to claim 7, wherein a number of consecutive copies are collected, the first copy having the same impulse response phase as desired. 9. The number p of base stations is not equal to a power of two, and the integer p ′ is determined to be greater than p and closest to p, which is a power of two, the processing being as if p The process according to claim 5, characterized in that there are 'base stations' and p'-p of these stations is virtual.